Study species and seed material
We focused on Biscutella didyma L., a winter annual Brassicaceae with Mediterranean and Irano-Turanian distribution (Feinbrun-Dothan & Danin 1998). This medium-sized, common and subordinate species was selected because it is one of few species occurring reliably across the entire rainfall gradient. As is typical for many annuals in our study region, B. didyma has short dispersal distances (Siewert & Tielbörger 2010) and low outcrossing rates (Boaz et al. 1990), making gene-flow among manipulated plots unlikely.
First, seeds of 100 B. didyma individuals (genotypes) per rainfall treatment and site were collected in spring 2012 by randomly picking twenty individuals within each 10m × 25m plot. Keeping ≥1.5m distance among genotypes and the large plot size ensured independence and capturing a representative fraction of the intrapopulation variance. One offspring per genotype was raised with regular watering for one generation in Tübingen (Germany) to reduce potential parental effects and amplify seed material. The growth period (Nov 2012 – May 2013) approximated the natural vegetation season and photoperiod, and greenhouse temperatures mirrored day temperatures in the field sites (15–20°C in winter, gradually warming to 25–30°C in spring). Wrapping flowering plants in light, transparent fabric (organza) ensured self-pollination and facilitated seed collection separately per mother. Second, of the resulting second generation, we randomly picked 40 genotypes per rainfall treatment and site (3×40 in SA, M; 40 in A, MM; altogether 320 genotypes) for the subsequent greenhouse experiment (detailed scheme: Appendix Fig. S1).